Research And Application Of Heart Sound Diagnosis System Based On Mobile Platform

Most countries aspire to deliver effective, safe, and affordable healthcare to theircitizens for the pursuit of healthy life. Today’s healthcare systems exist to lead large andgrowing budgets. The complexity and cost of hospital procedures seem to be one of themain reasons for the rising global medical service costs. And the resource constraints alsocause the difficulties of seeing the doctor. People were eager to see the doctor at home. Toovercome these obstacles around cost, access, and standards, healthcare system must behighly innovative and inventive to combine medical services and telecommunications,which is called m-Health (mobile health). It is simply a mobile medical system which iscombining the medical equipment with the popular mobile devices currently to achieve afast way of diagnosis and treatment.At present, the prevention and treatment of the chronic diseases has become one ofthe biggest problems in the world. And in theses chronic diseases, cardiovascular disease islet us more creeping. This is closely related to the level and style of our life.Cardiovascular disease (CVD) is a kind of disease with high incidence and strong sudden,which is the main reason of making the disease be a huge threat to human health. How toachieve the real-time monitoring and diagnosis of CVD has become the urgent thing of theworld, and it is also an important prerequisite for the successful realization of theprotection and treatment of CVD.The purpose of this paper is to use the Android smart-phone commonly used in our lifeto realize the preliminary diagnosis of heart sound signals, so as to realize the automaticdiagnosis and early warning of CVD. It is an advantage to realize the family care ofpatients with early CVD, and then to achieve the purpose of reducing the incidence ofCVD. In this paper, we used a new method—the state representation method (SRM) forheart sound identification, the experimental results showed that this method not only caneffectively identify the abnormal heart sounds and normal heart sounds, compared with thetraditional classification methods, such as artificial neural network (ANN) and supportvector machine (SVM), it can greatly shorten the calculation time as well. Considering the demands of cell phone users for the software experience, SRM is obviously more suitablefor the topic of the heart sound recognition based on mobile platform.The main research of this paper includes:(1) Preprocessing of heart sounds. Some traditional pretreatment methods were used inthe experiment, such as the mean, normalization and wavelet de-noising, etc. Through thispart of the preprocessing, the data quantization process in SRM algorithm can be saved.(2) Realize accurately segmentation of heart sound. Using multi-scale characteristicwaveform (MS-CW) and characteristic moment wave (CMW) to realize the accuratelysegmentation.(3) Extract heart sounds signal characteristic value from the multiple sides of timedomain, frequency domain and the time-frequency domain. The main methods includewavelet transform, MS-CW combined with CMW, and FFT. Select values which havesmall correlation through the analysis of the characteristics.(4) Recognize heart sound with a new method—SRM, which is used for recognizingthe normal and abnormal heart sounds. By building a model with the normal heart sounds,choosing an appropriate scale, to realize the recognition of heart sounds. In this paper,66groups of normal heart sounds and80groups of abnormal heart sounds were used to berecognized in the experiments. The result showed the accuracy of normal heart sounds’classification was100%, and the abnormal heart sounds reached93.8%.(5) Realize the heart sounds pre-diagnosis with Android platform. The whole softwarerealized several functions, including display of the heart sound file, save and the heartsounds diagnosis’ display and so on. Firstly realized the adjustment in the simulator ofEclipse, and then operated in the real machine.The results showed it could get the desired classification results using scale tool inSRM. Compared with traditional hospital cardiac auscultation, the system will realizefamily care of patients with CVD. In addition, compared with the traditional evaluationmethods, the system state can also describe the state of one person’s heart sound indifferent periods. It also provides a new way of thinking and method of prevention andtreatment of CVD in the future.